Sampling theorem
A signal can be rebuilt exactly from samples taken at twice its highest frequency.
Mathematics and information · Information theory · No. 128
- Electronic age
- Science
A Moscow engineer proved the theorem behind digital sound in 1933, but it drew little notice, and Shannon stated it again in 1949.
Chart: Illustration: Sampling theorem
Where it comes from
Chart: Map of where it comes from
Whittaker studied the mathematics in 1915 and Nyquist the telegraph limits in 1928. Kotelnikov gave a rigorous proof in a 1933 Russian conference report, but it gained little attention until Shannon restated it in 1949.
Three things to know
- Shannon called the result common knowledge in the communication art when he published it in 1949.
- Several people reached it independently: Whittaker in 1915, Kotelnikov in 1933, Raabe in 1939 and Shannon in 1949.
- Real systems sample somewhat faster than the minimum; the compact disc uses 44.1 kHz for sound that reaches about 20 kHz.
The journey
- 1 Smooth signalA sound or radio wave is a continuous curve
- 2 SamplesMeasure its height at evenly spaced instants
- 3 ProofKotelnikov proves nothing is lost if sampling is fast enough, 1933
- 4 SpreadShannon restates it for communication engineers, 1949
- 5 UseDigital telephony, recording, scopes and fax